why so hard to count Manhattan residential real estate sales?

lost some forest for some trees?
I got so caught up in micro-‘analysis’ yesterday (Mothra v. Godzilla, or the epic battle over Manhattan sales volume reports) that I did not get around to considering the factors that make it so hard to get accurate numbers about something as ‘simple’ as counting the number of coops and condos in Manhattan that closed between January 1 and March 31.

legitimate reasons for counting things not in city records
The Real Deal ran a piece yesterday that delves into some of the complexity. (Of course, once a firm starts making adjustments, they need to be consistent in doing so to have ‘the best’ numbers.)

• filing delays (close before end of quarter, filed with city after; can be a factor on either end but especially significant with December holiday delays in filings being recorded)
• managing agent data not (yet) in city records
• "non-legitimate" transfers recorded on ACRIS but not removed (close family transfers, or inter-corporate affiliate deals)

Let’s see how this can work, particularly for firms like Miller Samuel and Terra Holdings that have access to managing agent information for buildings being managed by friendly source (Douglas Elliman for MS; Halstead and Brown Harris for Terra) and information from agents in firms that do a lot of transactions (again, both firms), especially if those firms are on-site sales agents for new developments.

Douglas Elliman Property Management reports that X deals were due to close on Monday, March 31 in buildings they manage, or the new developments group at PruDE reports that XX deals were due to close in their buildings that same day. Miller Samuel’s report hit the press for publication on Wednesday, April 2, so the press had to have the report on Tuesday, April 1, so Miller Samuel had to cook-and-clean the report by (not later) than March 31. As these X and XX deals could not be recorded with the city by March 31 under any circumstances, no one other than principals and agents in the deals would know about them — and the PruDE organization is not sharing non-public information with competitors such as Terra. (Maybe this period for anticipated-deals-we-can’t-prove-closed is as long as a week.)

Miller Samuel is an appraisal firm. They may include deals that they have done a bank appraisal for, if they are confident that the closing  dates are fixed. (I have no idea if they do.)

Of course, the same is true for the Terra firms and their possible sources of information. They could (legitimately) include as Q1 transactions, deals that are closed-but-unknown-to-the-outside-world. (I think Terra has an affiliate or subsidiary relationship with an appraisal firm, but I am not so certain of this to identify who I think it is.)

So this is how firms can ‘over count’ deals in a quarter that are not yet public information. But this practice of adjustments creates data issues in future quarters.

data cops earn their dough
Again, I have no idea how well firms police their databases, but if a firm included X + XX deals as Q1 deals because they were told they had closed but to be recorded after March 31, they need to take those deals out of the Q2 report to be issued in 3 months, or they will have double-counted. For Q1 purposes, this may be especially a problem in over-counting deals that closed but to be recorded after December 31 if they already included them in Q407 reports. This may be especially difficult to adjust for if the deals expected to be closed at the end of the quarter did not close as planned, but ended up closing 6 or 8 weeks later — long enough afterwards that people forgot they had already been counted.

non-deal deals (for these purposes, at least)
Then, there are the phantom deals that are really reflected in city records as property transfers but are not arm’s length deals between unaffiliated buyers and sellers. I doubt these are a significant part of the ACRIS numbers in any quarter, but if an owner passes title in a condo to children or grandchildren for estate planning purposes, that does not reflect demand for Manhattan real estate, just family business. Same thing if a condo buyer who bought through a  corporation wants to change ownership to personal names (or vice versa). Or a coop owner who creates a residential real estate trust (with the coop’s OK, of course). Or (heaven forfend) a bank takes title to a condo in a foreclosure. You get the idea.

track, adjust, repeat
One would like to believe that careful reporting firms track this stuff carefully and understand what adjustments they make, measure over time whether the adjustments prove out, and have systems in place to avoid double-counting. Not to say anything negative about Terra, but Miller Samuel has been publicly issuing these reports for 14 years, they provide historical data for comparison (for an apples-to-apples comparison), they have a huge distribution to various kinds of market professionals interested in Manhattan real estate (so they can count on being ‘caught’ if their data prove unreliable) — not to mention that Jonathan Miller has been a voice in the wilderness for several years (at least) complaining about distortions in the mortgage approval and syndication process, so I have a high degree of confidence that he acts consistently in his reporting in this arena. (Again, nothing bad intended about Terra here; I just don’t know what i don’t know about them.)

a 3rd voice chimes in on The Number
Interesting that The Real Deal quoted a competing appraisal firm by name about the Q1 numbers. While Jeffrey Jackson, chairman of appraisal firm Mitchell, Maxwell & Jackson, did not reveal the numbers his firm has for Q108, he did tell The Real Deal "his own data agrees more closely with Miller Samuel’s lower figure of 2,282 first quarter sales". I wonder if Jackson would have made a public quote in the absence of Mothra publicly attacking Godzilla….

More Inside Baseball, I know. But it drives me nuts when I can’t reconcile ‘simple’ numbers.

Especially when the numbers reported are so specific — 2,282 from Miller Samuel, not 2,281 or 2,283. As the Terra guy told his troops (as quoted in The Real Deal):

the numbers are not presented "as hard fact, but as an indication of the scope of the report."